1. NVIDIA GB10 Grace Blackwell Superchip with 1 PFLOP AI Performance

The MSI EdgeXpert is built around the NVIDIA GB10 Grace Blackwell Superchip, combining a Blackwell GPU with a 20-core Arm CPU (10× Cortex-X925 + 10× Cortex-A725). Delivering up to 1 PFLOP (1000 AI TOPS FP4) of AI compute performance, it is designed for local AI development, inference, model fine-tuning, data science, and machine learning workloads that would traditionally require data-center hardware.

2. Massive 128GB Unified Memory for Large AI Models

Unlike traditional desktops that separate CPU and GPU memory, the EdgeXpert uses 128GB of unified LPDDR5X memory connected through NVLink-C2C, creating a coherent memory pool accessible by both processor and GPU. This architecture allows developers to run AI models with up to 200 billion parameters locally, while two linked systems can scale to support models of up to 405 billion parameters.

3. Compact Personal AI Supercomputer with Enterprise Connectivity

Despite delivering data-center-class AI performance, the EdgeXpert fits into a 1.19-litre chassis measuring just 151 × 151 × 52 mm. The system includes 4TB self-encrypting NVMe storage, 10Gb Ethernet, ConnectX-7 Smart NIC, Wi-Fi 7, Bluetooth, USB-C connectivity, and NVIDIA DGX OS. This enables seamless migration of AI workloads from desktop development environments to DGX Cloud or larger NVIDIA-accelerated infrastructures.